jp-codereview

Automated code review issue generation works by analyzing phase artifacts like prd.md, tasks.yaml, and prior reviews to identify quality gaps. It then writes structured issues directly to a review cycle YAML file linked to related tasks.

A review cycle YAML file is a structured output that records code review issues for a specific phase. It is generated as reviews/cycle/NNN.yaml, capturing identified gaps and linking them directly to related tasks from your tasks.yaml file.

To avoid duplicate code review issues across rounds, the process checks an existing fixes.yaml file before generating new output. It respects previously recorded fixes to ensure only new or unresolved gaps are written to the next review cycle.

Yes, you can scope an automated code review to changed files by specifying the scope when invoking the review task. It uses the full or changed scope parameter to determine which files within the phase directory to analyze.

Generating structured code review issues requires phase artifacts including prd.md, tasks.yaml, and prior review rounds. The system validates these inputs in the specified phase directory before creating the output YAML file.

Code review issues are linked to tasks.yaml entries to maintain traceability between identified quality gaps and their originating requirements. This ensures each generated issue in the cycle file is mapped to a related task for clear tracking.